Questions on Value Averaging

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
tmcnulty1982
Posts: 6
Joined: Thu Jul 28, 2011 10:30 pm

Questions on Value Averaging

Post by tmcnulty1982 » Thu Nov 10, 2011 7:44 pm

I just finished reading Edleson's book on Value Averaging (the '06 version) and I have a few questions that hopefully will spark some interest:

1) I'm not sure I understand the theory that value averaging (VA) works on quarterly periods better than it does on monthly periods (pp. 160-4). It would seem that he is saying that the market tends to overreact in the short term, and my understanding of VA is that it works by capitalizing on those short term swings in the market. Indeed, there would be no advantage over dollar cost averaging (DCA) in a completely uniform investment vehicle. Doesn't this mean that the more short term swings (the more you can sell high and buy low) the better? Any thoughts on if or why there is a limit to the frequency with which it makes sense to VA?

2) I'm also struggling to decide if it makes more sense to VA a single fund (Total Stock Mkt Idx, Total International Mkt Idx, Inter-Term Bond Mkt Idx) or an entire portfolio made up of these funds? My understanding of the role of bonds in a portfolio is that they tend to be inversely correlated to stocks, so they inherently help smooth out the bumps. If VA benefits from higher price variations, it seems that adding bonds as part of the VA portfolio (rather than a Short-Term Bond Mkt Idx side fund) may reduce the benefits of VA. On the other hand, if they were VA'ed separately (possibly sharing a single side fund), they may demonstrate more variability individually.

3) Lastly, what is the prescribed way to "get out" of a VA scenario once you've hit your mark? Is there a sanctioned method to reduce risk exposure as you near the mark in case of any unexpected down turns? One way to do this would be to accept a bond fund as part of the portfolio and gradually increase the percentage of bonds held, but this would have an affect on expected growth and require regular adjustments to your value path (which is not necessarily bad, just more complex and harder to predict the goal amount). Another way might be to "reverse value average," where you gradually sell your investments to maintain a set reduction in value over time. Does anyone have any experience with these or other methods of reducing risk as you near the portfolio end-goal?

Thanks in advance.
Last edited by tmcnulty1982 on Sun Jan 11, 2015 3:47 pm, edited 1 time in total.

User avatar
bertilak
Posts: 6922
Joined: Tue Aug 02, 2011 5:23 pm
Location: East of the Pecos, West of the Mississippi

Re: Questions on Value Averaging

Post by bertilak » Fri Nov 11, 2011 11:23 am

I have not read the book, but do have some idea of what value averaging is. Here is my take on your questions/comments...

1. Longer periods will smooth out short-term noise. No sense buying "low" when even lower is just around the corner. You also want to minimize transaction costs.

2. Multiple assets: You should have an asset allocation (percent of each asset). Value averaging involves setting an ever-increasing target. Simply have that increase be the same for all asset classes (to preserve the overall AA) and invest in each as necessary to maintain your goal.

3. "get out" scenario. Not sure what it means to "hit your mark" as it is an ever-evolving target. So nothing to get out of. If you intend to start taking withdrawals, I suppose you could "reverse VA". This would be a periodic, constant percentage, withdrawal. Keeping that to below about 4% gives you some hope of preserving principal. Most withdrawal schemes are already exactly that.
May neither drought nor rain nor blizzard disturb the joy juice in your gizzard. -- Squire Omar Barker (aka S.O.B.), the Cowboy Poet

Topic Author
tmcnulty1982
Posts: 6
Joined: Thu Jul 28, 2011 10:30 pm

Re: Questions on Value Averaging

Post by tmcnulty1982 » Sat Nov 12, 2011 6:32 pm

1) That makes sense, but again, isn't the "noise" what's being taken advantage of by VA? I.e., isn't there an equal or better chance that the asset will increase in value a month later, rather than decrease in value? Obviously looking at the data he does show that quarterly is better than monthly, but I'm trying to understand the theory rather than rely on past experiences. Oh, and in my case there are no transaction costs.

2) Good call. I like the idea of VA'ing each asset class separately, but using the same expected growth for each to maintain the same AA. I was thinking about doing VA for the total value of the portfolio and AA inside of that, but your method of thinking about it seems simpler to me.

3) That the target is ever evolving is also good advice. Do you know of a VA formula that takes into account an R that decreases over time? It would be nice to plot a realistic value path based on a portfolio that will be expected to return gradually less and less each year as the % of bonds increases.

User avatar
LadyGeek
Site Admin
Posts: 57819
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Questions on Value Averaging

Post by LadyGeek » Sat Nov 12, 2011 7:00 pm

The wiki has some background info on value averaging: Dollar cost averaging (Value averaging)

Take a good look at the spreadsheet. You need to be very careful in a down market; as you'll need to find extra funds to keep this technique working. For this reason, value averaging is not recommended for new investors.

(I don't want to discuss the merits of dollar cost vs. value averaging, just point out some tutorial information.)
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

umfundi
Posts: 3361
Joined: Tue Jun 07, 2011 5:26 pm

Re: Questions on Value Averaging

Post by umfundi » Sat Nov 12, 2011 9:39 pm

tmcnulty1982 wrote:I just finished reading Edleson's book on Value Averaging (the '06 version) and I have a few questions that hopefully will spark some interest:

1) I'm not sure I understand the theory that value averaging (VA) works on quarterly periods better than it does on monthly periods (pp. 160-4). It would seem that he is saying that the market tends to overreact in the short term, and my understanding of VA is that it works by capitalizing on those short term swings in the market. Indeed, there would be no advantage over dollar cost averaging (DCA) in a completely uniform investment vehicle. Doesn't this mean that the more short term swings (the more you can sell high and buy low) the better? Any thoughts on if or why there is a limit to the frequency with which it makes sense to VA?

2) I'm also struggling to decide if it makes more sense to VA a single fund (Total Stock Mkt Idx, Total International Mkt Idx, Inter-Term Bond Mkt Idx) or an entire portfolio made up of these funds? My understanding of the role of bonds in a portfolio is that they tend to be inversely correlated to stocks, so they inherently help smooth out the bumps. If VA benefits from higher price variations, it seems that adding bonds as part of the VA portfolio (rather than a Short-Term Bond Mkt Idx side fund) may reduce the benefits of VA. On the other hand, if they were VA'ed separately (possibly sharing a single side fund), they may demonstrate more variability individually.

3) Lastly, what is the prescribed way to "get out" of a VA scenario once you've hit your mark? Is there a sanctioned method to reduce risk exposure as you near the mark in case of any unexpected down turns? One way to do this would be to accept a bond fund as part of the portfolio and gradually increase the percentage of bonds held, but this would have an affect on expected growth and require regular adjustments to your value path (which is not necessarily bad, just more complex and harder to predict the goal amount). Another way might be to "reverse value average," where you gradually sell your investments to maintain a set reduction in value over time. Does anyone have any experience with these or other methods of reducing risk as you near the portfolio end-goal?

Thanks in advance,
Tobias
Tobias,

I hope I don't get in to too much trouble here.

Value averaging is a variant of Dollar Cost Averaging, which is a non-optimal strategy.
the market tends to overreact in the short term
This is the idea of "momentum", which is market timing and a bet on market inefficiencies that none of us mere mortals can hope to exploit.

If you accept that the central tenet of the Boglehead philosophy is to establish and maintain an asset allocation, DCA and VA are incompatible with that philosophy.

Which is not to say you shouldn't do it. But, you should have a consistent philosophy.

Keith
Déjà Vu is not a prediction

Topic Author
tmcnulty1982
Posts: 6
Joined: Thu Jul 28, 2011 10:30 pm

Re: Questions on Value Averaging

Post by tmcnulty1982 » Sat Nov 12, 2011 10:22 pm

Hi LadyGeek and umfundi:

Thanks for your replies. I have a basic understanding of value averaging and realize that it may involve coming up short on cash in a down year, or sitting on loads of cash in good years (which may indeed be inefficient). These are valid concerns and not to be overlooked.

That said, I don't want to stir up another conversation about which is better, AA, DCA, VA, etc., and the numerous ways they can be combined, confused, or conflated. I.e., right now at least, I'm less interested in practical advice about what to do with my money today and more interested in understanding the theory behind what a "pure value averaging" strategy would like like, from planning through execution and completion.

I hope that helps. Thanks for any input you might have!

Tobias

AndroAsc
Posts: 1219
Joined: Sat Nov 21, 2009 7:39 am

Re: Questions on Value Averaging

Post by AndroAsc » Sun Nov 13, 2011 12:59 am

tmcnulty1982 wrote:Thanks for your replies. I have a basic understanding of value averaging and realize that it may involve coming up short on cash in a down year, or sitting on loads of cash in good years (which may indeed be inefficient). These are valid concerns and not to be overlooked.
I think most of the forumers here practice DCA and not VA. I once posted the same questions as you did, and it spiraled down to senseless argument with other people on the semantics of bookkeeping asset allocations. I do not have a good solution to these issues related to VA, although I would be happy to discuss them.

Probably the easier problem to solve is "sitting on loads of cash in good years". One could have a threshold for maximum amount of cash. For e.g., if the excess "cash equivalents" exceeds 10% of your entire portfolio, you would put them into your desired asset allocation the next time you invest. The trouble with this approach is that you may just happen to be at the peak of the market and invested this extra chunk of cash just before a crash. Another downside is that you might not have enough cash to top up your VA path when the market tanks, which brings us to the next issue "coming up short on cash in down year". This is one that I do not have a good solution to. In fact, I faced this problem in my last investment. These two fundamental VA problems to my knowledge have no solution.

I only have a partial solution to the VA problem. First, I create a VA path that will give me the desired lump sum at retirement, and I adopt a "target sum" philosophy. This means that my goal is to get that target sum when I retire, as opposed to wanting to have 6% p.a. For the "excess cash" problem, I do not think I will address it directly, cause there is too much that can go wrong if you start tweaking around (like in the example I provided above). Yes, it is inefficient if so much cash is uninvested, but the worst case scenario here is that you would have potentially gotten more money at retirement. Now, compare this to the scenario where you try to minimize the excess cash and find yourself coming up short during a crash. This would have a greater downside that your final target sum at retirement would not be met. In other words, the potential loss of returns by sitting in cash should be less of a concern than the potential inability to meet the target sum at retirement.

AndroAsc
Posts: 1219
Joined: Sat Nov 21, 2009 7:39 am

Re: Questions on Value Averaging

Post by AndroAsc » Sun Nov 13, 2011 1:11 am

In terms of practical execution of VA:
1) Create a VA path based on two considerations - how much I can practically save AND the need to reach the target sum at retirement.
2) Extract out the monthly savings rate for the VA path and save/invest that amount.
3) Using my asset allocation, calculate how much assets I should have in the various funds (stocks, bonds, etc.)
4) At every time point (quarterly), top up my portfolio to most closely replicate VA path value and their respective allocation in the various funds.
5) Excess cash stays as cash in bull runs. I may do a form of "tactical asset allocation" in really obvious situations For e.g., if the market is currently at 50% from its peak and I'm still sitting on a crap load of cash... in this situation I may just dump more cash into my portfolio.
6) Repeat step (1) whenever your economic situation changes (e.g. increased salary that would increase your savings)

Obviously, if you are VA purists, you would have noted minor changes to the original VA plan. I put forth that this implementation has the following advantages:
1) You have a "fixed" amount to save every month, making it compatible with IRA/401k where there is max contribution per month. In the original VA implementation, the monthly savings was a variable value and I think that's not practically implementable.
2) You have an asset allocation that suits your risk appetite. Original VA implementation says nothing about asset allocation.

AndroAsc
Posts: 1219
Joined: Sat Nov 21, 2009 7:39 am

Re: Questions on Value Averaging

Post by AndroAsc » Sun Nov 13, 2011 1:23 am

tmcnulty1982 wrote:1) That makes sense, but again, isn't the "noise" what's being taken advantage of by VA? I.e., isn't there an equal or better chance that the asset will increase in value a month later, rather than decrease in value? Obviously looking at the data he does show that quarterly is better than monthly, but I'm trying to understand the theory rather than rely on past experiences. Oh, and in my case there are no transaction costs.
Yes, VA only "works" because of volatility. The more volatile the market, the more VA will outperform DCA. I have no explanation on the difference between monthly and quarterly VA, are the differences significant enough to bother with in the first place?
tmcnulty1982 wrote:2) Good call. I like the idea of VA'ing each asset class separately, but using the same expected growth for each to maintain the same AA. I was thinking about doing VA for the total value of the portfolio and AA inside of that, but your method of thinking about it seems simpler to me.

3) That the target is ever evolving is also good advice. Do you know of a VA formula that takes into account an R that decreases over time? It would be nice to plot a realistic value path based on a portfolio that will be expected to return gradually less and less each year as the % of bonds increases.
I also take a similar approach, but I have different expected returns for bonds and stocks, and based on my current asset allocation, my VA path expected return will be a composite of the expected return of bonds and stocks. For e.g., stocks 7% p.a., bonds 3% p.a., 50/50 allocation, VA expected returns will be 5% p.a. With this approach, you can simulate a decreasing expected return of VA path, as you adjust your asset allocation to be more conservative with time.

tpm871
Posts: 149
Joined: Sun Sep 12, 2010 10:58 am

Re: Questions on Value Averaging

Post by tpm871 » Sun Nov 13, 2011 2:57 am

1) I'm not sure I understand the theory that value averaging (VA) works on quarterly periods better than it does on monthly periods (pp. 160-4). It would seem that he is saying that the market tends to overreact in the short term, and my understanding of VA is that it works by capitalizing on those short term swings in the market. Indeed, there would be no advantage over dollar cost averaging (DCA) in a completely uniform investment vehicle. Doesn't this mean that the more short term swings (the more you can sell high and buy low) the better? Any thoughts on if or why there is a limit to the frequency with which it makes sense to VA?
I've done value averaging since 2007. At first, I used quarterly periods as the book suggested, but I became a little discouraged that I missed good buying and selling points in doing so. But then I discovered Opportunistic Rebalancing, and using tolerance bands:

http://www.tdainstitutional.com/pdf/Opp ... yanani.pdf

So now I'm still using VA, but I've incorporated opportunistic rebalancing as well. I look at how I'm doing twice per month, and only make adjustments if values go above or below the bands that I've set. But what I'm doing is a hybrid of the two approaches -- my target for each period is determined by my value path (i.e., as opposed to the current percentage of my portfolio) but I check more frequently and use bands to react more quickly to buy/sell opportunities. The bands allow for momentum in the market, to reduce the odds of buying or selling prematurely. It also has the effect of reducing transactions.
2) I'm also struggling to decide if it makes more sense to VA a single fund (Total Stock Mkt Idx, Total International Mkt Idx, Inter-Term Bond Mkt Idx) or an entire portfolio made up of these funds? My understanding of the role of bonds in a portfolio is that they tend to be inversely correlated to stocks, so they inherently help smooth out the bumps. If VA benefits from higher price variations, it seems that adding bonds as part of the VA portfolio (rather than a Short-Term Bond Mkt Idx side fund) may reduce the benefits of VA. On the other hand, if they were VA'ed separately (possibly sharing a single side fund), they may demonstrate more variability individually.
I have an asset allocation of a bunch of different asset classes, including bonds. I have a separate value path for each one. However, since the higher level strategy is to manage risk of my entire portfolio via an AA, my path for each asset class is determined in part by where I think the whole portfolio will be at the end of the year. What this generally means is that I add more to the lower performing asset classes during the year (e.g., emerging markets earlier this year) and less to the higher performing asset classes -- or even trim back by selling a little (e.g., I trimmed back on my European equities towards the end of October, after a sudden spike).

Usually, you'll buy bonds more frequently since they tend to not keep pace with equities, but the opposite is true at times (e.g., during this summer). The purpose of bonds is to have something to draw cash from when you need to buy equities when they are low (although it takes a lot of guts to do this).

I look at this approach as a way of systematically buying low and selling high. It's a lot easier to stick to it if you have a plan in advance like VA.
3) Lastly, what is the prescribed way to "get out" of a VA scenario once you've hit your mark? Is there a sanctioned method to reduce risk exposure as you near the mark in case of any unexpected down turns? One way to do this would be to accept a bond fund as part of the portfolio and gradually increase the percentage of bonds held, but this would have an affect on expected growth and require regular adjustments to your value path (which is not necessarily bad, just more complex and harder to predict the goal amount). Another way might be to "reverse value average," where you gradually sell your investments to maintain a set reduction in value over time. Does anyone have any experience with these or other methods of reducing risk as you near the portfolio end-goal?
I've done this slightly, but not a lot. Like many Bogleheads, I do the "your age in bonds" approach to gradually changing my AA each year (in my case, I use my age minus 10). So every year I shift 1% from my AA targets from equities to bonds. But this doesn't mean immediately selling anything -- it's just a slight adjustment to my VA path. Basically, I "rebaseline" my plan with a slightly new AA and new starting values each year.

---

And as for the issue of running out of cash when the market goes down: yes, it can happen. It's only happened to me once, though. In late January 2009, I ran out of cash. Not a big deal. The market will either recover (e.g., like it did strarting in March 2009), or you'll reach a rebaseline point and can start drawing money from bonds.

As for the issue of carrying too much cash when the market is doing well: I've had several occasions when this has happened, only to be followed by deep dips. I was much better off buying durin those dips than I would have been being fully invested at all times. I got some good deals with my excess cash during August and September this year, for example.

AndroAsc
Posts: 1219
Joined: Sat Nov 21, 2009 7:39 am

Re: Questions on Value Averaging

Post by AndroAsc » Sun Nov 13, 2011 11:00 am

tpm871 wrote:I've done value averaging since 2007. At first, I used quarterly periods as the book suggested, but I became a little discouraged that I missed good buying and selling points in doing so. But then I discovered Opportunistic Rebalancing, and using tolerance bands:

http://www.tdainstitutional.com/pdf/Opp ... yanani.pdf

So now I'm still using VA, but I've incorporated opportunistic rebalancing as well. I look at how I'm doing twice per month, and only make adjustments if values go above or below the bands that I've set. But what I'm doing is a hybrid of the two approaches -- my target for each period is determined by my value path (i.e., as opposed to the current percentage of my portfolio) but I check more frequently and use bands to react more quickly to buy/sell opportunities. The bands allow for momentum in the market, to reduce the odds of buying or selling prematurely. It also has the effect of reducing transactions.
Interesting, I've read about opportunistic rebalancing before but never formally incorporated into my investment strategy. What tolerance band do you use? 20%?
tpm871 wrote:And as for the issue of running out of cash when the market goes down: yes, it can happen. It's only happened to me once, though. In late January 2009, I ran out of cash. Not a big deal. The market will either recover (e.g., like it did strarting in March 2009), or you'll reach a rebaseline point and can start drawing money from bonds.

As for the issue of carrying too much cash when the market is doing well: I've had several occasions when this has happened, only to be followed by deep dips. I was much better off buying durin those dips than I would have been being fully invested at all times. I got some good deals with my excess cash during August and September this year, for example.
That's a pretty good experience so far, but I guess since you started at 2007, you would naturally have more cash than someone who started in 2010 (like me). What values did you use in the expected returns of your VA path?

User avatar
LadyGeek
Site Admin
Posts: 57819
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Questions on Value Averaging

Post by LadyGeek » Sun Nov 13, 2011 11:47 am

Good discussion; I added this thread to the wiki.

Wiki article link: Dollar cost averaging (under External links, Value averaging)
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

tpm871
Posts: 149
Joined: Sun Sep 12, 2010 10:58 am

Re: Questions on Value Averaging

Post by tpm871 » Sun Nov 13, 2011 1:51 pm

Interesting, I've read about opportunistic rebalancing before but never formally incorporated into my investment strategy. What tolerance band do you use? 20%?
I'm doing something a bit different here too. My thought was that different asset classes have different levels of volatility, so the bands for each should be different to best catch their local minimums and maximums. For example, emerging markets needs wider bands than intermediate term treasuries.

I also thought that it makes sense to have different upper and lower bands, since advances and declines may behave differently. So what I ended up doing was measuring all of the advances and declines for the past decade, and using the median advance and decline to set the band size for each asset class. Here's some examples:

Code: Select all

Asset    Lower band    Upper band
Sm val.        2.9%      15.0%
Em mkts.       9.1%      14.8%
Int treasury.  2.1%       2.7%
Europe index.  5.4%       6.3%
I buy far more frequently that I sell -- selling is a fairly rare event.
That's a pretty good experience so far, but I guess since you started at 2007, you would naturally have more cash than someone who started in 2010 (like me). What values did you use in the expected returns of your VA path?
The amount of cash varies based on market performance. As the market goes down, you become more fully invested. As it goes up, you build up more of a cash reserve. But note that everytime you rebaseline, your targets are reset. So if you end the year with a lot of cash, your value path for the following year will be more aggressive. In my case, I ended last year with a lot of cash. Since my path became more aggressive, a lot of that cash was absorbed doing August and September in keeping my equity funds on their value path. At the moment, I don't have nearly as much in cash and my equities are starting to hit their upper bands.

I read somewhere that having an expected return below 9% was non-optimal for VA. I currently use 11% for my portfolio as a whole. Note that you'll almost never be right in what you pick, but that's ok. It's far more common to have the market perform significantly above or below the average than right at the average, so I don't worry too much about the expected return that I choose in planning the value path.

Post Reply